211 datasets found

The dataset used in the paper is a large-scale corpus of movie reviews from the Socher et al. (2013) dataset.

Cross-topic Argument Mining from Heterogeneous Sources.

Semeval-2016 Task 6: Detecting stance in tweets.

Stance detection aims to identify whether the author of a text is in favor of, against, or neutral to a given target. The main challenge of this task comes two-fold: few-shot...

The dataset is a combination of 3 singular datasets: A Million News Headlines, Fake and real news, Getting Real about Fake News.

The Rotten Tomatoes dataset has 5331 positive and 5331 negative review sentences.

The dataset used for toxicity and stereotype mitigation task, which consists of 25 thousand examples of positive and negative movie reviews.

The dataset used in the paper is a concatenation of the original Harry Potter books and synthetic discussions, blog posts, and wiki-like entries about the books.

The dataset used in the paper is the Equity Evaluation Corpus (EEC) for emotion prediction, which contains a balanced dataset of sentences with emotions.

The dataset used for the proposed Joint Multi-Domain Learning for Automatic Short Answer Grading.

The IMDb review dataset is used for positive generation task.

The dataset used in the LightDXML paper for extreme multi-label classification.

The Pile dataset is a large-scale dataset containing 800GB of text data.

The LM-Extraction benchmark is derived from The Pile (Gao et al., 2020) dataset, which contains 15,000 pairs of prefixes and suffixes derived from The Pile dataset (Gao et al.,...

The dataset used in the paper is the TREC05 spam corpus, which contains 39,999 real ham and 52,790 spam emails.

The dataset consists of 108 news headlines, 72 of which are true and 36 of which are false.

Deep learning-based text classification models need abundant labeled data to obtain competitive performance. To tackle this, multiple researches try to use data augmentation to...

The gisette dataset is a collection of 20,000 text documents, each containing a single sentence.

The epsilon dataset is a collection of 50,000 text documents, each containing a single sentence.

The paper proposes dual-sparse regularized randomized reduction methods for classification. The dataset used in the paper is the RCV1-binary dataset.